Artificial Intelligence By Example

Designing the architecture of the data flow graph

First import TensorFlow with the following code:

import tensorflow as tf

Then use the following code to give a name of the scope of each object you want to display in one line. Write your variable or object on the second line and include it in the TensorBoard summary as an image (histograms, scalars, and other summaries are possible), as shown in the following code:

with tf.name_scope("Input"): x_ = tf.placeholder(tf.float32, shape=[4,2], name = 'x-input-predicates') tf.summary.image('input predicates x', x_, 10)

name provides a label for the objects in the graph. scope is a way to group the names to simplify the visual representation of your architecture.

Repeat that with all the elements ...